Databricks Time Travel: Tutorial on Recovering Delta Tables

If you’ve accidentally written incorrect data to a Delta Lake table in Databricks, you can potentially recover the table using Delta Lake’s time travel feature. Time travel allows you to access previous versions of a Delta table, making it possible to revert to a point before the incorrect data was written.

Table of contents

1. Databricks time travel to recover a Delta Lake table
   1. Identify the Correct Version
   2. Restore the Table:
   3. Verify the Data

Databricks time travel to recover a Delta Lake table

Identify the Correct Version

Determine the version of the Delta Lake table that contains the correct data. You can view the history of changes to a Delta table using the DESCRIBE HISTORY command in Databricks.

DESCRIBE HISTORY EMP

This command will show you a list of all the versions of the Delta table along with timestamps and actions performed.

Restore the Table:

Once you’ve identified the correct version, you can restore the Delta Lake table to that version using the RESTORE command.

RESTORE EMP TO VERSION AS OF 2

Verify the Data

After restoring the table, verify that the correct data has been recovered by querying the table.

SELECT * FROM EMP

Ensure that the data matches your expectations and that the incorrect data has been replaced with the correct version.

Conclusion

By using Delta Lake’s time travel feature, you can go back to an earlier version of a Delta table and restore it to a state where the wrong data wasn’t there. But keep in mind that time travel is affected by retention policies and older versions of a Delta table might be deleted automatically based on those policies. So it’s important to act quickly to recover the table before the version you want is removed.